Ciena Offers Up Virtual WAN, Tears Down Data Center Walls
May 23, 2012
Virtualization and cloud computing are hot topics at EMC World--but how do you deliver speed and data center resiliency, as well as scalability, in virtual and cloud environments? Ciena is focusing on tackling all three issues with its demonstrations at this week's conference in Las Vegas.
Jim Morin, product line director, managed services and enterprise, at Ciena, says the company's technology platform for a virtual WAN will facilitate the deployment of ultra-low-latency stock market applications and ease the distribution of synchronous virtual machine architectures. In the long run, it will allow the development of bandwidth-on-demand services for data-intensive cloud applications.
- Client Windows Migration: Expert Tips for Application Readiness
- Thwart off Application-Based Security Exploits: Protect Against Zero-Day Attacks, Malware, Advanced Persistent Threats
- Best Practices for Security and Compliance with Amazon Web Services
- Why a New Business Model is Needed for SSL Certificates
- State of Cloud 2011: Time for Process Maturation
- SaaS 2011: Adoption Soars, Yet Deployment Concerns Linger
Ciena is showcasing at EMC World its Data Center Without Walls: a live vMotion application operating over a distance of 100 km, and distributed storage between VPLEX clusters in a simulated cloud environment. The system allows for dynamic network scalability for the optical networks used in WANs, to create a more reliable, flexible and affordable shared networking infrastructure. "The transportation through Ciena's system and the protection and resiliency built into the network allow organizations to extend the full latency tolerance and synchronous replication of a system like EMC VPLEX," says Morin, who's also deputy commissioner of the TechAmerica Foundation's CLOUD2 group.
The virtual machine concept has simplified the development and deployment of new applications by abstracting many of the complexities in modern software systems. Likewise, virtual storage systems have simplified the management of distributed databases and file systems. However, the management of network resources has typically operated in a domain outside the purview or control of application developers. In traditional network systems, a developer has to wait hours, or even days, to provision the bandwidth for a new application. These changes can be further complicated by security requirements such as those from the Payment Card Industry or the Health Insurance Portability and Accountability Act.
Virtual LAN technology can help overcome many of these challenges within an organization and, to some extent, over larger networks. However, over larger distances there are important distinctions between the ways that optical and electrical networks are routed and provisioned. This creates limits on the pricing structures and dynamic flexibility that carriers can offer. Consequently, organizations need to buy enough bandwidth to satisfy peak demand.
Communication latency restricts the pace at which applications can keep in step, and increases with distance and congestion. The speed of light adds about 1 ms of latency per 100 km. In addition, processing delays associated with Internet routing add other, non-deterministic delays that can grow substantially as networks reach peak capacity. A virtual machine replication platform like VPLEX can tolerate a delay of up to 5 ms.
Industries such as finance have seen measurable advantages by minimizing the time to gather market data and execute algorithmic trades. According to Morin, these organizations will benefit from more streamlined data transfer systems that take advantage of the lowest possible latency in the box. In synchronous applications, each virtual machine waits on updates from the others before it can execute the next step in processing. "These apps are looking for a clean Layer 1 transfer with a deterministic performance," Morin says.
Organizations generally address the bottleneck by overprovisioning bandwidth, Morin adds. In some cases, it can take a telco 60 days to change the network, so organizations buy enough to handle their expected growth. "This adds up to inefficiencies because we are rarely transmitting at the full capacity of the network," explains Morin. "Performance on demand can react to workloads programmatically to allow the hypervisor to adjust the capacity, and when done to move the network back down to a steady state so you are only paying for premium bandwidth when you need it."
Ciena is working with software developers and carriers to integrate the network management hypervisor into management, billing and security services. "It is a complete ecosystem demonstrating how you can dial into the system, request more capacity--and the bill shows that," Morin says.
IDC analyst Cindy Borovick says she expects to see more innovation on the WAN in support of cloud computing and hybrid cloud offerings.